2
\$\begingroup\$

I'm exploring the possibility of using TI's TS5A3160 analog switch to implement a "buffer" for a square wave. The idea is, to use a square wave with a low driving capability as the control signal of the switch, which will have as its inputs VCC and GND. So, output of switch will commutate between VCC and GND, with a duty cycle and period given by the control square wave, and switching times (rise/fall) given by the switch timing capabilities.

Input square wave will be 10 MHz, required rise and fall for output are less than 10 ns, so switch capabilities are OK. What is concerning me is the make-before-break time (MBB)

a enter image description here

If I understand correctly, for a period of time less than 15 ns, inputs will be shorted (I suppose connected with an impedance in the order of Ron, 1 ohm). In this case inputs are VCC and GND. In all application notes and references I found, MBB concerns are related to output glitches. However in this case I'm not concerned about that, but about the short causing damage to the power supply, or the chip.

I tend to intuitively believe that such a "short" shortcircuit wouldn't be damaging. However I'm lacking a SPICE model of the switch to simulate it, and I can't think of ways of evaluating the impact of this short. Should I be concerned about this? Are there ways to mitigate this problem? Is there some sort of simulation I could run (without the switch model) to estimate the magnitude of this problem?

\$\endgroup\$

1 Answer 1

2
\$\begingroup\$

Connecting the output to VCC or GND is what every digital output does. So you can replace this analog switch with a logic buffer, e.g., SN74LVC1G34.

In a CMOS device, the current flowing from VCC to GND through the two partially open transistors is called shoot-through current, and it definitely happens. (In theory, it would be possible to add a break-before-make delay, but this would increase the transmission delay, and logic devices are optimized for speed.) The devices themselves are designed to handle this, and the resulting current spikes on the power supply are why all logic devices need a nearby decoupling capacitor.

If you want to avoid that current (and are not using a very high-frequency signal), use a switch with break-before-make. If you really want to use the TS5A3160 (although tMBB can be larger than 10 ns), ensure that you have a decoupling capacitor (which is alreay needed by the digital switching done through the control input).

\$\endgroup\$
6
  • \$\begingroup\$ Thanks for you answer. Regarding using a logical BUFFER instead of an analog MUX, I was initially discouraged to take this approach since I couldn't find buffers which output more than 100 mA, and they were all limited to 5V operation (this ended up applying to MUXES too). Regarding using a decoupling cap to compensate this problem, I intuitively thought it should work, but however I tried some SPICE simulations (with a coarse model for the switch) and it didn't work. The spikes in current in digital circuits you mention, are they due to shoot-through current or to charging/discharging caps? \$\endgroup\$
    – MPA95
    May 2 at 15:16
  • \$\begingroup\$ While looking for components I also noticed that break-before-make were always slower than MBB. It is good to know that is a technological limitation, thanks. And a question, on the first line of the second paragraph, when you wrote "the current flowing from VCC to GND through the two partially open transistors", did you actually mean "partially CLOSED transistors"? I'm assuming you did \$\endgroup\$
    – MPA95
    May 2 at 15:19
  • 1
    \$\begingroup\$ You did not mention the voltage and current you want, but look at gate drivers like the UCC27517. \$\endgroup\$
    – CL.
    May 2 at 18:06
  • 1
    \$\begingroup\$ The transistors are neither fully open nor fully closed, but partially open, which is the same as partically closed. \$\endgroup\$
    – CL.
    May 2 at 18:07
  • \$\begingroup\$ thanks, that driver looks OK, however I will need a high side switch also, since my application can not work with a pull-up resistor. And I looked into BUFFERS but unfortunately they do not output VCC... VOH tends to be lower than VCC, and most datasheets inform it in facts decreases with increasing current... \$\endgroup\$
    – MPA95
    May 2 at 23:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.